pi-namespace-patch 0.85.1 → 0.86.1-namespace.1

package/docs/models.md

@@ -9,6 +9,7 @@ Add custom providers and models (Ollama, vLLM, LM Studio, proxies) via `~/.pi/ag
 - [Supported APIs](#supported-apis)
 - [Provider Configuration](#provider-configuration)
 - [Model Configuration](#model-configuration)
+- [Prompt Cache Lifetimes](#prompt-cache-lifetimes)
 - [Overriding Built-in Providers](#overriding-built-in-providers)
 - [Per-model Overrides](#per-model-overrides)
 - [Anthropic Messages Compatibility](#anthropic-messages-compatibility)
@@ -208,6 +209,7 @@ If your command is slow, expensive, rate-limited, or should keep using a previou
 | `maxTokens` | No | `16384` | Maximum output tokens |
 | `samplingParams` | No | omitted | Sampling parameters merged verbatim into every request body (see below) |
 | `cost` | No | all zeros | Per-million-token rates with optional request-wide input pricing tiers |
+| `promptCache` | No | omitted | Best-effort prompt cache lifetime in seconds per retention tier (see below) |
 | `compat` | No | provider `compat` | Provider compatibility overrides. Merged with provider-level `compat` when both are set. |
 
 A cost tier supplies a complete alternate rate set and applies to the full request when total input usage (`input + cacheRead + cacheWrite`) exceeds `inputTokensAbove`. When multiple tiers match, the highest threshold wins.
@@ -236,6 +238,19 @@ Current behavior:
 - `/model`, `--list-models`, and the interactive footer display entries by model `id`.
 - The configured `name` is used for model matching and secondary model detail text. It does not replace the footer/status-bar model id.
 
+### Prompt Cache Lifetimes
+
+`promptCache` states how long the provider keeps a prompt cache entry alive for each retention tier pi can request (`short` is the default tier; `long` is used when `PI_CACHE_RETENTION=long`). Values are seconds and are estimates: providers publish ranges, so pick the conservative end.
+
+```json
+{
+  "id": "claude-sonnet-5",
+  "promptCache": { "short": 300, "long": 3600 }
+}
+```
+
+The built-in catalog fills this in for direct Anthropic (5 min / 1 h). Other providers, including direct OpenAI, have no built-in lifetime until their cache-expiry and replay behavior has been validated for warming. A model without a value for the tier a request used is never warmed; custom models and provider overrides can opt in when the backing cache behavior is known. See [Cache Warming](settings.md#cache-warming).
+
 ### Sampling Parameters
 
 `samplingParams` is a free-form object merged verbatim into every request body for the model, after the fields pi sets itself, so its keys win. Use it to send sampling parameters pi does not model — including server-specific ones like llama.cpp's `min_p` or vLLM's `top_k`:
@@ -359,7 +374,23 @@ Use `modelOverrides` to customize built-in models and matching extension-registe
 }
 ```
 
-`modelOverrides` supports these fields per model: `name`, `reasoning`, `thinkingLevelMap`, `input`, `cost` (partial), `contextWindow`, `maxTokens`, `samplingParams` (merged per key), `headers`, `compat`.
+`modelOverrides` supports these fields per model: `name`, `reasoning`, `thinkingLevelMap`, `input`, `cost` (partial), `promptCache` (merged per tier), `contextWindow`, `maxTokens`, `samplingParams` (merged per key), `headers`, `compat`.
+
+Use a `promptCache` override to enable cache warming through a proxy whose backing cache you know, for example OpenRouter routed to Anthropic:
+
+```json
+{
+  "providers": {
+    "openrouter": {
+      "modelOverrides": {
+        "anthropic/claude-sonnet-4": {
+          "promptCache": { "short": 300 }
+        }
+      }
+    }
+  }
+}
+```
 
 Direct OpenAI GPT-5.6 Sol, Terra, and Luna default to a `272000` context window so requests remain within OpenAI's short-context pricing tier. To opt into OpenAI's 1.05M context window, increase it for each model you use:
 
@@ -435,6 +466,7 @@ Built-in Anthropic models enable `supportsStrictTools` in their model metadata.
 | `supportsMidConvoEffort` | Whether the exact Claude model transport supports per-turn effort system messages and thinking binding controls. Pi persists native effort levels and always sends `drop_block` when enabled. Default: `false`. |
 | `allowEmptySignature` | Whether to replay empty thinking signatures as `signature: ""` instead of converting thinking to text. Default: `false`. |
 | `supportsStrictTools` | Whether the provider accepts strict JSON-schema tool definitions. Default: `false`; built-in Anthropic models enable it in generated metadata. |
+| `allowedFallbackModels` | Up to three server-side fallback models, each with `provider`, `model`, and complete `cost` metadata. An empty array disables fallback. |
 
 ## OpenAI Compatibility
 
@@ -481,7 +513,6 @@ For providers with partial OpenAI compatibility, use the `compat` field.
 | `sessionAffinityFormat` | For `openai-completions` and `openai-responses`, the session-affinity header format: `openai` sends `session_id`/`x-client-request-id` (completions also `x-session-affinity`), `openai-nosession` omits the underscore-containing `session_id` header, `openrouter` sends `x-session-id`. Does not affect the `prompt_cache_key` body param. Default: auto-detected. |
 | `supportsStrictMode` | Whether the provider accepts strict JSON-schema function tool definitions. Defaults depend on the API; built-in OpenAI models carry explicit capability metadata. |
 | `supportsOpenAIGrammarTools` | Whether OpenAI-compatible APIs emit custom Lark/regex grammar tools. When `false`, grammar-constrained tools fall back to normal function tools. Default: `false`; the built-in model catalog enables it for GPT-5+ models on OpenAI, OpenAI Codex, Azure OpenAI, GitHub Copilot, opencode, and Cloudflare AI Gateway. |
-| `deferredToolsMode` | Use provider-specific deferred tool serialization. Currently only `"kimi"` is supported for Kimi's OpenAI-compatible Chat Completions format. |
 | `supportsLongCacheRetention` | Whether the provider accepts long cache retention when cache retention is `long`: `prompt_cache_options.ttl: "30m"` for GPT-5.6+ Responses models, `prompt_cache_retention: "24h"` for earlier OpenAI models, or `cache_control.ttl: "1h"` when `cacheControlFormat` is `anthropic`. Default: `true`. |
 | `openRouterRouting` | OpenRouter provider routing preferences. This object is sent as-is in the `provider` field of the [OpenRouter API request](https://openrouter.ai/docs/guides/routing/provider-selection). |
 | `vercelGatewayRouting` | Vercel AI Gateway routing config for provider selection (`only`, `order`) |

package/docs/providers.md

@@ -20,6 +20,7 @@ Use `/login` in interactive mode, then select a provider:
 - Claude Pro/Max
 - GitHub Copilot
 - xAI (Grok/X subscription)
+- Meta (Muse subscription)
 - OpenRouter (OAuth-minted API key billed from OpenRouter credits)
 - Radius
 
@@ -44,6 +45,12 @@ Anthropic subscription auth is active for Claude Pro/Max accounts. Third-party h
 - Run `/login xai`, then select **Use a subscription**
 - `XAI_API_KEY` remains available through **Use an API key**
 
+### Meta (Muse subscription)
+
+- Run `/login meta`, then select **Sign in with Meta** to open the device authorization flow
+- The login mints a Model API key that is re-minted automatically about once a day
+- `META_API_KEY` remains available through **Use an API key**
+
 ### OpenRouter
 
 - Run `/login openrouter`, then select **Sign in with OpenRouter** to open the OpenRouter PKCE authorization flow
@@ -53,7 +60,7 @@ Anthropic subscription auth is active for Claude Pro/Max accounts. Third-party h
 
 ### Radius
 
-Radius is a dynamic `pi-messages` gateway. `/login radius` stores OAuth tokens in `auth.json`; the gateway catalog is refreshed independently and cached in `models-store.json`. Custom Radius gateways can be declared in `models.json` with `"oauth": "radius"` and a gateway `baseUrl`.
+Radius is a `pi-messages` gateway. Pi ships the public Radius model catalog for immediate and offline model lookup, then overlays it with the effective gateway catalog after authentication. `/login radius` stores OAuth tokens in `auth.json`; refreshed catalogs are cached in `models-store.json`. Custom Radius gateways can be declared in `models.json` with `"oauth": "radius"` and a gateway `baseUrl`; they do not inherit the public `radius.pi.dev` catalog.
 
 ## API Keys
 
@@ -94,6 +101,7 @@ pi
 | Together AI | `TOGETHER_API_KEY` | `together` |
 | Baseten | `BASETEN_API_KEY` | `baseten` |
 | Kimi For Coding | `KIMI_API_KEY` | `kimi-coding` |
+| Meta | `META_API_KEY` | `meta` |
 | MiniMax | `MINIMAX_API_KEY` | `minimax` |
 | MiniMax (China) | `MINIMAX_CN_API_KEY` | `minimax-cn` |
 | Qwen Token Plan (existing catalog) | `QWEN_TOKEN_PLAN_API_KEY` | `qwen-token-plan` |
@@ -104,7 +112,7 @@ pi
 | Xiaomi MiMo Token Plan (Amsterdam) | `XIAOMI_TOKEN_PLAN_AMS_API_KEY` | `xiaomi-token-plan-ams` |
 | Xiaomi MiMo Token Plan (Singapore) | `XIAOMI_TOKEN_PLAN_SGP_API_KEY` | `xiaomi-token-plan-sgp` |
 
-Reference for environment variables and `auth.json` keys: [`const envMap`](https://github.com/earendil-works/pi-mono/blob/main/packages/ai/src/env-api-keys.ts) in [`packages/ai/src/env-api-keys.ts`](https://github.com/earendil-works/pi-mono/blob/main/packages/ai/src/env-api-keys.ts).
+Reference for environment variables and `auth.json` keys: [`const envMap`](https://github.com/earendil-works/pi/blob/main/packages/ai/src/env-api-keys.ts) in [`packages/ai/src/env-api-keys.ts`](https://github.com/earendil-works/pi/blob/main/packages/ai/src/env-api-keys.ts).
 
 #### Auth File
 

package/docs/sdk.md

@@ -111,6 +111,8 @@ interface AgentSession {
 }
 ```
 
+`session.navigateTree()` rejects while an agent response, manual or automatic compaction, or another tree navigation is active, even with `summarize: false`. It does not queue navigation or return `{ cancelled: true }` for these conflicts. Wait for the active operation to finish (for example, with `await session.waitForIdle()`) and retry. Rejection leaves the active branch unchanged.
+
 Session replacement APIs such as new-session, resume, fork, and import live on `AgentSessionRuntime`, not on `AgentSession`.
 
 ### createAgentSessionRuntime() and AgentSessionRuntime
@@ -244,8 +246,8 @@ const state = session.agent.state;
 // state.messages: AgentMessage[] - conversation history
 // state.model: Model - current model
 // state.thinkingLevel: ThinkingLevel - current thinking level
-// state.systemPrompt: string - system prompt
-// state.tools: AgentTool[] - available tools
+// state.systemPrompt: string - read-only, replayed from the transcript's system messages
+// state.tools: AgentTool[] - executable tools; changes are declared to the model before the next request
 // state.streamingMessage?: AgentMessage - current partial assistant message
 // state.errorMessage?: string - latest assistant error
 

package/docs/security.md

@@ -54,6 +54,6 @@ If you bind-mount a host workspace read/write, writes from inside the container
 
 ## Reporting Security Issues
 
-To report a security issue, follow the repository [Security Policy](https://github.com/earendil-works/pi-mono/blob/main/SECURITY.md). Do not open a public issue for security-sensitive reports.
+To report a security issue, follow the repository [Security Policy](https://github.com/earendil-works/pi/blob/main/SECURITY.md). Do not open a public issue for security-sensitive reports.
 
 Expected local-agent behavior, lack of a built-in sandbox, prompt injection from untrusted content, and behavior of user-installed extensions or skills are generally outside the security boundary unless the report demonstrates a real privilege-boundary bypass or shows how pi grants access that the local user did not already have.

package/docs/session-format.md

@@ -5,10 +5,10 @@ Sessions are stored as JSONL (JSON Lines) files. Each line is a JSON object with
 ## File Location
 
 ```
-~/.pi/agent/sessions/--<path>--/<timestamp>_<uuid>.jsonl
+~/.pi/agent/sessions/--<path>--/<timestamp>_<session-id>.jsonl
 ```
 
-Where `<path>` is the working directory with `/` replaced by `-`.
+By default, `<session-id>` is a UUID. Callers can supply a custom ID through the SDK or `--session-id`. For `<path>`, Pi removes the leading path separator and replaces `/`, `\\`, and `:` with `-`.
 
 ## Deleting Sessions
 
@@ -28,11 +28,11 @@ Existing sessions are automatically migrated to the current version (v3) when lo
 
 ## Source Files
 
-Source on GitHub ([pi-mono](https://github.com/earendil-works/pi-mono)):
-- [`packages/coding-agent/src/core/session-manager.ts`](https://github.com/earendil-works/pi-mono/blob/main/packages/coding-agent/src/core/session-manager.ts) - Session entry types and SessionManager
-- [`packages/coding-agent/src/core/messages.ts`](https://github.com/earendil-works/pi-mono/blob/main/packages/coding-agent/src/core/messages.ts) - Extended message types (BashExecutionMessage, CustomMessage, etc.)
-- [`packages/ai/src/types.ts`](https://github.com/earendil-works/pi-mono/blob/main/packages/ai/src/types.ts) - Base message types (UserMessage, AssistantMessage, ToolResultMessage)
-- [`packages/agent/src/types.ts`](https://github.com/earendil-works/pi-mono/blob/main/packages/agent/src/types.ts) - AgentMessage union type
+Source on GitHub ([pi](https://github.com/earendil-works/pi)):
+- [`packages/coding-agent/src/core/session-manager.ts`](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/src/core/session-manager.ts) - Session entry types and SessionManager
+- [`packages/coding-agent/src/core/messages.ts`](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/src/core/messages.ts) - Extended message types (BashExecutionMessage, CustomMessage, etc.)
+- [`packages/ai/src/types.ts`](https://github.com/earendil-works/pi/blob/main/packages/ai/src/types.ts) - Base message types (UserMessage, AssistantMessage, ToolResultMessage)
+- [`packages/agent/src/types.ts`](https://github.com/earendil-works/pi/blob/main/packages/agent/src/types.ts) - AgentMessage union type
 
 For TypeScript definitions in your project, inspect `node_modules/@earendil-works/pi-coding-agent/dist/` and `node_modules/@earendil-works/pi-ai/dist/`.
 
@@ -48,6 +48,7 @@ Messages contain arrays of typed content blocks:
 interface TextContent {
   type: "text";
   text: string;
+  textSignature?: string;
 }
 
 interface ImageContent {
@@ -59,6 +60,8 @@ interface ImageContent {
 interface ThinkingContent {
   type: "thinking";
   thinking: string;
+  thinkingSignature?: string;
+  redacted?: boolean;
 }
 
 interface ToolCall {
@@ -66,12 +69,22 @@ interface ToolCall {
   id: string;
   name: string;
   arguments: Record<string, any>;
+  thoughtSignature?: string;
+  namespace?: string;
 }
 ```
 
 ### Base Message Types (from pi-ai)
 
 ```typescript
+interface SystemMessage {
+  role: "system";
+  content: string | TextContent[];
+  toolsAdded?: Tool[];
+  toolsRemoved?: Array<{ name: string }>;
+  timestamp: number;  // Unix ms
+}
+
 interface UserMessage {
   role: "user";
   content: string | (TextContent | ImageContent)[];
@@ -84,9 +97,16 @@ interface AssistantMessage {
   api: string;
   provider: string;
   model: string;
+  responseModel?: string;
+  responseId?: string;
+  providerThinkingLevel?: string;
+  diagnostics?: AssistantMessageDiagnostic[];
   usage: Usage;
-  stopReason: "stop" | "length" | "toolUse" | "error" | "aborted";
+  stopReason: "pending" | "stop" | "length" | "toolUse" | "error" | "aborted" | "deferred";
+  deferred?: DeferredHandle;
   errorMessage?: string;
+  rawStopReason?: string;
+  endTurn?: boolean;
   timestamp: number;
 }
 
@@ -106,6 +126,8 @@ interface Usage {
   output: number;
   cacheRead: number;
   cacheWrite: number;
+  cacheWrite1h?: number;
+  reasoning?: number;
   totalTokens: number;
   cost: {
     input: number;
@@ -117,7 +139,7 @@ interface Usage {
 }
 ```
 
-The exported pi-ai `StopReason` type also includes `"pending"`, but that value is reserved for partial messages in streaming events. Terminal `done`/`error` messages replace it with a completion reason before pi persists the assistant message, so `"pending"` should never appear in session JSONL.
+`"pending"` is reserved for partial messages in streaming events. Terminal events replace it with a completion reason before Pi persists the assistant message, so `"pending"` should never appear in session JSONL. `"deferred"` is a terminal reason for a provider response that will complete later; its `deferred` handle contains the provider data needed to retrieve that response.
 
 ### Extended Message Types (from pi-coding-agent)
 
@@ -146,7 +168,7 @@ interface CustomMessage {
 interface BranchSummaryMessage {
   role: "branchSummary";
   summary: string;
-  fromId: string;                // Entry we branched from
+  fromId: string | null;         // Previous leaf whose abandoned path was summarized
   timestamp: number;
 }
 
@@ -162,6 +184,7 @@ interface CompactionSummaryMessage {
 
 ```typescript
 type AgentMessage =
+  | SystemMessage
   | UserMessage
   | AssistantMessage
   | ToolResultMessage
@@ -178,8 +201,8 @@ All entries (except `SessionHeader`) extend `SessionEntryBase`:
 ```typescript
 interface SessionEntryBase {
   type: string;
-  id: string;           // 8-char hex ID
-  parentId: string | null;  // Parent entry ID (null for first entry)
+  id: string;           // Usually an 8-char hex ID; may fall back to a full UUID
+  parentId: string | null;  // Parent entry ID (null for a root entry)
   timestamp: string;    // ISO timestamp
 }
 ```
@@ -202,12 +225,19 @@ For sessions with a parent (created via `/fork`, `/clone`, or `newSession({ pare
 
 ### SessionMessageEntry
 
-A message in the conversation. The `message` field contains an `AgentMessage`.
+A message in the conversation. The `message` field contains an `AgentMessage`. System messages carry the prompt and tool loadout: the first request of a session persists one with every prompt section and tool declaration, and later changes persist as system messages that patch `sections` by name (`null` removes one) and list `toolsAdded`/`toolsRemoved`. Replaying them in order yields the current prompt and tools; there is no separate prompt state entry.
+
+```json
+{"type":"message","id":"a0b1c2d3","parentId":null,"timestamp":"2024-12-03T14:00:00.000Z","message":{"role":"system","content":"","sections":{"preamble":"You are an expert coding assistant...","tools":"<tools>\n- read: ...\n</tools>","cwd":"/project"},"toolsAdded":[{"name":"read","description":"...","parameters":{}}],"timestamp":1733234400000}}
+{"type":"message","id":"d4e5f6g7","parentId":"c3d4e5f6","timestamp":"2024-12-03T14:04:00.000Z","message":{"role":"system","content":"","sections":{"skills":"<skills>...</skills>"},"toolsRemoved":[{"name":"write"}],"timestamp":1733234640000}}
+```
+
+Sessions created before system messages existed have no leading system message; the first request declares the current prompt as a later system message, which replays the same way.
 
 ```json
-{"type":"message","id":"a1b2c3d4","parentId":"prev1234","timestamp":"2024-12-03T14:00:01.000Z","message":{"role":"user","content":"Hello"}}
-{"type":"message","id":"b2c3d4e5","parentId":"a1b2c3d4","timestamp":"2024-12-03T14:00:02.000Z","message":{"role":"assistant","content":[{"type":"text","text":"Hi!"}],"provider":"anthropic","model":"claude-sonnet-4-5","usage":{...},"stopReason":"stop"}}
-{"type":"message","id":"c3d4e5f6","parentId":"b2c3d4e5","timestamp":"2024-12-03T14:00:03.000Z","message":{"role":"toolResult","toolCallId":"call_123","toolName":"bash","content":[{"type":"text","text":"output"}],"isError":false}}
+{"type":"message","id":"a1b2c3d4","parentId":"prev1234","timestamp":"2024-12-03T14:00:01.000Z","message":{"role":"user","content":"Hello","timestamp":1733234401000}}
+{"type":"message","id":"b2c3d4e5","parentId":"a1b2c3d4","timestamp":"2024-12-03T14:00:02.000Z","message":{"role":"assistant","content":[{"type":"text","text":"Hi!"}],"api":"anthropic-messages","provider":"anthropic","model":"claude-sonnet-4-5","usage":{...},"stopReason":"stop","timestamp":1733234402000}}
+{"type":"message","id":"c3d4e5f6","parentId":"b2c3d4e5","timestamp":"2024-12-03T14:00:03.000Z","message":{"role":"toolResult","toolCallId":"call_123","toolName":"bash","content":[{"type":"text","text":"output"}],"isError":false,"timestamp":1733234403000}}
 ```
 
 ### ModelChangeEntry
@@ -226,26 +256,31 @@ Emitted when the user changes the thinking/reasoning level.
 {"type":"thinking_level_change","id":"e5f6g7h8","parentId":"d4e5f6g7","timestamp":"2024-12-03T14:06:00.000Z","thinkingLevel":"high"}
 ```
 
-### CompactionEntry
+### UsageEntry
 
-Created when context is compacted. Stores a summary of earlier messages.
+Records model-attributed usage that is not an assistant message and does not participate in LLM context. `kind` is an arbitrary string identifying the operation; for example, cache warming uses `"cache_warm"`.
 
 ```json
-{"type":"compaction","id":"f6g7h8i9","parentId":"e5f6g7h8","timestamp":"2024-12-03T14:10:00.000Z","summary":"User discussed X, Y, Z...","firstKeptEntryId":"c3d4e5f6","tokensBefore":50000}
+{"type":"usage","id":"f6g7h8i9","parentId":"e5f6g7h8","timestamp":"2024-12-03T14:08:00.000Z","kind":"cache_warm","provider":"anthropic","model":"claude-sonnet-4-5","usage":{"input":0,"output":0,"cacheRead":50000,"cacheWrite":0,"totalTokens":50000,"cost":{"input":0,"output":0,"cacheRead":0.015,"cacheWrite":0,"total":0.015}}}
 ```
 
-Newer harness-generated compactions embed the retained post-compaction context directly on the entry, instead of `firstKeptEntryId`:
+Usage entries contribute to session token and cost totals. Pi hides them from the conversation tree. Consumers should treat unknown `kind` values as normal usage rather than rejecting them.
+
+### CompactionEntry
+
+Created when context is compacted. Stores a summary of earlier messages and a complete system prompt/tool checkpoint.
 
 ```json
-{"type":"compaction","id":"f6g7h8i9","parentId":"e5f6g7h8","timestamp":"2024-12-03T14:10:00.000Z","summary":"User discussed X, Y, Z...","tokensBefore":50000,"retainedTail":[{"role":"user","content":"latest request"},{"role":"assistant","content":[{"type":"text","text":"latest reply"}],"provider":"anthropic","model":"claude-sonnet-4-5","usage":{...},"stopReason":"stop"}]}
+{"type":"compaction","id":"f6g7h8i9","parentId":"e5f6g7h8","timestamp":"2024-12-03T14:10:00.000Z","summary":"User discussed X, Y, Z...","firstKeptEntryId":"c3d4e5f6","tokensBefore":50000,"systemMessage":{"role":"system","content":"You are a coding assistant.","toolsAdded":[],"timestamp":1733235000000}}
 ```
 
+`firstKeptEntryId` is required. It identifies the first entry retained from before the compaction entry. When rebuilding context, Pi replaces older summarized entries with the compaction summary and keeps the range beginning at this entry.
+
 Optional fields:
+- `systemMessage`: The replayed prompt sections and tool declarations at the compaction boundary; it becomes the leading system message of the compacted context, and system messages among the kept entries are dropped in its favor. It is absent on older session entries.
 - `usage`: LLM usage from generating the summary; included in session token and cost totals
-- `retainedTail`: Materialized `AgentMessage[]` kept after compaction. This is optional only for backward compatibility with older sessions. Newer harness-generated compactions include it so we can rebuild context from this checkpoint without walking older entries before the compaction entry.
 - `details`: Implementation-specific data (e.g., `{ readFiles: string[], modifiedFiles: string[] }` for default, or custom data for extensions)
 - `fromHook`: `true` if generated by an extension, `false`/`undefined` if pi-generated (legacy field name)
-- `firstKeptEntryId`: for compatibility with old entry format.
 
 ### BranchSummaryEntry
 
@@ -255,6 +290,8 @@ Created when switching branches via `/tree` with an LLM generated summary of the
 {"type":"branch_summary","id":"g7h8i9j0","parentId":"a1b2c3d4","timestamp":"2024-12-03T14:15:00.000Z","fromId":"f6g7h8i9","summary":"Branch explored approach A..."}
 ```
 
+`parentId` is the entry from which the new branch continues. `fromId` is the previous leaf whose abandoned path was summarized.
+
 Optional fields:
 - `usage`: LLM usage from generating the summary; included in session token and cost totals
 - `details`: File tracking data (`{ readFiles: string[], modifiedFiles: string[] }`) for default, or custom data for extensions
@@ -305,11 +342,12 @@ The session name is displayed in the session selector (`/resume`) instead of the
 
 ## Tree Structure
 
-Entries form a tree:
-- First entry has `parentId: null`
-- Each subsequent entry points to its parent via `parentId`
+Entries normally form one tree, but navigation APIs can create multiple roots:
+- A root entry has `parentId: null`; the first entry is initially the root
+- Each non-root entry points to its parent via `parentId`
 - Branching creates new children from an earlier entry
 - The "leaf" is the current position in the tree
+- Calling `resetLeaf()` or `branchWithSummary(null, ...)` allows a later entry to become another root
 
 ```
 [user msg] ─── [assistant] ─── [user msg] ─── [assistant] ─┬─ [user msg] ← current leaf
@@ -322,11 +360,10 @@ Entries form a tree:
 `buildContextEntries()` walks from the current leaf to the root, producing the active entry list while honoring compaction:
 
 1. Collects all entries on the path
-2. If a `CompactionEntry` is on the path:
+2. If one or more `CompactionEntry` values are on the path, uses the latest one:
    - Includes the compaction entry first
-   - If `retainedTail` is present, it acts as a self-contained checkpoint and entries after the compaction are included
-   - Otherwise entries from `firstKeptEntryId` to the compaction are included
-   - Then entries after compaction are included
+   - Includes non-system entries from `firstKeptEntryId` up to, but not including, the compaction entry
+   - Includes entries after the compaction entry
 3. Preserves non-message entries in the selected range so interactive mode can render them
 
 `buildSessionContext()` builds on that entry list to produce the message list for the LLM:
@@ -334,12 +371,12 @@ Entries form a tree:
 1. Extracts current model and thinking level settings from the full path
 2. Converts selected entries to messages:
    - `message` -> stored `AgentMessage`
-   - `compaction` -> `compactionSummary` plus `retainedTail` when present
+   - `compaction` -> complete system checkpoint followed by `compactionSummary`
    - `branch_summary` -> `branchSummary`
    - `custom_message` -> `CustomMessage`
-   - `custom` -> no context message
+   - `usage` and `custom` -> no context message
 
-This makes newer compactions act like self-contained checkpoints. `retainedTail` is optional only so older sessions that only store `firstKeptEntryId` continue to load correctly.
+The compaction summary replaces entries before `firstKeptEntryId`. Pre-compaction system messages are folded into the complete checkpoint rather than replayed from the retained range. Retained non-system entries and all entries after the compaction remain available to the LLM.
 
 ## Parsing Example
 
@@ -364,6 +401,9 @@ for (const line of lines) {
     case "branch_summary":
       console.log(`[${entry.id}] Branch from ${entry.fromId}`);
       break;
+    case "usage":
+      console.log(`[${entry.id}] Usage (${entry.kind}): ${entry.usage.totalTokens} tokens`);
+      break;
     case "custom":
       console.log(`[${entry.id}] Custom (${entry.customType}): ${JSON.stringify(entry.data)}`);
       break;
@@ -388,18 +428,19 @@ for (const line of lines) {
 Key methods for working with sessions programmatically.
 
 ### Static Creation Methods
-- `SessionManager.create(cwd, sessionDir?)` - New session
-- `SessionManager.open(path, sessionDir?)` - Open existing session file
+- `SessionManager.create(cwd, sessionDir?, options?)` - New session; `options` can set `id` and `parentSession`
+- `SessionManager.open(path, sessionDir?, cwdOverride?)` - Open existing session file
 - `SessionManager.continueRecent(cwd, sessionDir?)` - Continue most recent or create new
-- `SessionManager.inMemory(cwd?)` - No file persistence
-- `SessionManager.forkFrom(sourcePath, targetCwd, sessionDir?)` - Fork session from another project
+- `SessionManager.inMemory(cwd?, options?, entries?)` - No file persistence, optionally initialized from entries
+- `SessionManager.forkFrom(sourcePath, targetCwd, sessionDir?, options?)` - Fork session from another project
 
 ### Static Listing Methods
 - `SessionManager.list(cwd, sessionDir?, onProgress?)` - List sessions for a directory
 - `SessionManager.listAll(onProgress?)` - List all sessions across all projects
+- `SessionManager.listAll(sessionDir?, onProgress?)` - List sessions from a custom session root
 
 ### Instance Methods - Session Management
-- `newSession(options?)` - Start a new session (options: `{ parentSession?: string }`)
+- `newSession(options?)` - Start a new session (options: `{ id?: string, parentSession?: string }`)
 - `setSessionFile(path)` - Switch to a different session file
 - `createBranchedSession(leafId)` - Extract branch to new session file
 
@@ -407,7 +448,8 @@ Key methods for working with sessions programmatically.
 - `appendMessage(message)` - Add message
 - `appendThinkingLevelChange(level)` - Record thinking change
 - `appendModelChange(provider, modelId)` - Record model change
-- `appendCompaction(summary, firstKeptEntryId, tokensBefore, details?, fromHook?)` - Add compaction
+- `appendUsage(kind, provider, model, usage)` - Record model-attributed usage outside the conversation
+- `appendCompaction(summary, firstKeptEntryId, tokensBefore, details?, fromHook?, usage?)` - Add compaction
 - `appendCustomEntry(customType, data?)` - Extension state (not in context)
 - `appendSessionInfo(name)` - Set session display name
 - `appendCustomMessageEntry(customType, content, display, details?)` - Extension message (in context)
@@ -423,7 +465,7 @@ Key methods for working with sessions programmatically.
 - `getLabel(id)` - Get label for entry
 - `branch(entryId)` - Move leaf to earlier entry
 - `resetLeaf()` - Reset leaf to null (before any entries)
-- `branchWithSummary(entryId, summary, details?, fromHook?)` - Branch with context summary
+- `branchWithSummary(entryId, summary, details?, fromHook?, usage?)` - Branch with context summary; `entryId` may be `null` to branch from the root
 
 ### Instance Methods - Context & Info
 - `buildContextEntries()` - Get active branch entries with compaction applied

package/docs/sessions.md

@@ -33,6 +33,7 @@ For the JSONL file format and SessionManager API, see [Session Format](session-f
 | `/compact [prompt]` | Summarize older context; see [Compaction](compaction.md) |
 | `/export [file]` | Export session to HTML |
 | `/share` | Upload as private GitHub gist with shareable HTML link |
+| `/bug [description]` | Report a bug to the Pi developers; see [Reporting Bugs](#reporting-bugs) |
 
 ## Resuming and Deleting Sessions
 
@@ -138,6 +139,32 @@ When prompted, choose one of:
 
 See [Compaction](compaction.md) for branch summarization internals and extension hooks.
 
+## Reporting Bugs
+
+`/bug [description]` collects a bug report for the Pi developers. The report is not shared publicly. The dialog asks for an optional description and whether to include the session transcript. If you decline the transcript, pi offers to have the current model write a summary of what went wrong instead; the transcript is sent to your provider with your credentials, and only the summary is attached.
+
+The last step chooses where the report goes:
+
+- **Upload Report** sends it to the Pi developers through `radius.pi.dev`. No login is required; if you are logged into Radius, the report is attributed to your account so the developers can follow up. If the upload fails, pi offers to export the zip instead.
+- **Export as Zip** writes a zip archive to the current directory. Attach it to an issue or send it to the developers yourself.
+
+Both contain the same files:
+
+| File | Content |
+|------|---------|
+| `report.json` | pi version, runtime, OS, terminal, current model and provider configuration, loaded extensions, and settings. API keys, header values, URL credentials, and the analytics tracking id are never included. |
+| `diagnostics.json` | Provider and runtime error diagnostics attached to assistant messages across the whole session (failed or aborted turns, retries, error messages), plus any recorded crashes. Always included; message content is not. |
+| `session.jsonl` | The current branch of the session, only when you chose to include it. It contains file contents and command output read during the session. |
+| `summary.md` | The model-written summary, only when you chose to generate one. |
+
+Each report has a UUID. pi shows it after upload or export and records it in the session as a `pi.bug-report` entry so you can refer to it later.
+
+Set `PI_RADIUS_GATEWAY` to upload to a different Radius deployment.
+
+### Crashes
+
+When pi exits because of an uncaught exception or a fatal runtime error, it stores the error message and stack trace in `~/.pi/agent/crashes.json` (the newest five). The next interactive start shows a warning once; running `/bug` attaches the stored crashes to `diagnostics.json` and removes the file after the report is uploaded or exported. Resume the crashed session with `pi -r` first if you want the transcript in the report.
+
 ## Session Format
 
 Session files are JSONL and contain message entries, model changes, thinking-level changes, labels, compactions, branch summaries, and extension entries.

package/docs/settings.md

@@ -32,8 +32,31 @@ Use `/trust` in interactive mode to save a project trust decision for future ses
 | `defaultThinkingLevel` | string | - | Startup thinking level (saved with Ctrl+S in `/thinking`, or edited manually): `"off"`, `"minimal"`, `"low"`, `"medium"`, `"high"`, `"xhigh"`, `"max"` |
 | `modelThinkingLevels` | object | - | Per-model startup thinking levels keyed by `"provider/modelId"`; configure from `/settings` → Default thinking level per model or edit manually |
 | `hideThinkingBlock` | boolean | `false` | Hide thinking blocks in output |
-| `showCacheMissNotices` | boolean | `false` | Show transcript notices for significant prompt-cache misses, compaction or branch-summary usage, and provider recovery diagnostics such as dropped Anthropic thinking blocks |
+| `showCacheMissNotices` | boolean | `false` | Show transcript notices for significant prompt-cache misses, successful cache-warming usage, compaction or branch-summary usage, and provider recovery diagnostics such as dropped Anthropic thinking blocks |
 | `thinkingBudgets` | object | - | Custom token budgets per thinking level. Anthropic, Google, and Bedrock use these natively. OpenAI-compatible models use them when `compat.thinkingTokenBudgetField` (or `supportsThinkingTokenBudget`) is set. |
+| `cacheWarming` | string | `"streaming"` | Prompt cache-warming mode: `"off"`, `"streaming"`, or `"idle"`. Global setting only. |
+
+#### Cache Warming
+
+Providers drop a prompt cache entry after a period of inactivity, so the first request after a pause pays full input price again. Cache warming re-sends the last request with a one-token output budget shortly before expiry:
+
+- `"off"` disables warming.
+- `"streaming"` protects expensive prefixes during long tool executions and stops as soon as the agent settles.
+- `"idle"` also considers refreshes while waiting for your next prompt, using a fixed 15% continuation probability measured from real usage.
+
+```json
+{
+  "cacheWarming": "idle"
+}
+```
+
+A refresh is sent only when the expected avoided cache-miss cost, minus the cost of the refresh, leaves at least $0.05 of expected savings. Active agent runs use 100% continuation probability. `/session` shows the next decision, continuation probability, expected savings, threshold, and estimated costs. When cache miss notices are enabled, each successful refresh appears in the transcript with its cost; notices identify extension overrides.
+
+Warming stops when the context changes (model switch, compaction, branch navigation). Idle warming stops no later than 30 minutes after the last real provider request; warming during an active agent run stops after 60 minutes. Extensions can override each decision through the [`cache_warming_decision`](extensions.md#cache_warming_decision) event.
+
+Each refresh is billed as a cache read of the full context plus one output token. Usage and cost show up in session totals but never enter model context. Pi schedules candidates at 90% of the cache lifetime while leaving at least ten seconds before expiry.
+
+Warming needs a known cache lifetime for the model and the retention tier the request used (`short`, or `long` with `PI_CACHE_RETENTION=long`). The built-in catalog carries lifetimes for direct Anthropic; custom models and other providers can declare theirs with `promptCache` in `models.json` (see [Prompt Cache Lifetimes](models.md#prompt-cache-lifetimes)). Claude models that use budget-based rather than adaptive thinking are skipped while thinking is on, because Anthropic derives the thinking budget from `max_tokens` and keys the message cache on it, so a one-token request cannot reproduce the entry.
 
 #### thinkingBudgets
 
@@ -118,6 +141,7 @@ Set `PI_SKIP_VERSION_CHECK=1` to disable the Pi version update check. Use `--off
 | `compaction.enabled` | boolean | `true` | Enable auto-compaction |
 | `compaction.reserveTokens` | number | `16384` | Tokens reserved for LLM response |
 | `compaction.keepRecentTokens` | number | `20000` | Recent tokens to keep (not summarized) |
+| `compaction.modelOverrides` | object | - | Per-model `reserveTokens` and `keepRecentTokens` overrides keyed by exact `"provider/modelId"` |
 
 ```json
 {
@@ -129,6 +153,37 @@ Set `PI_SKIP_VERSION_CHECK=1` to disable the Pi version update check. Use `--off
 }
 ```
 
+#### Per-model compaction overrides
+
+```json
+{
+  "compaction": {
+    "enabled": true,
+    "reserveTokens": 16384,
+    "keepRecentTokens": 20000,
+    "modelOverrides": {
+      "some-provider/big-model": {
+        "reserveTokens": 400000
+      },
+      "local/small-model": {
+        "reserveTokens": 2048,
+        "keepRecentTokens": 4096
+      }
+    }
+  }
+}
+```
+
+Keys match exact, case-sensitive `provider/modelId` values, not names or glob patterns. Model IDs may contain slashes (for example, `openrouter/anthropic/claude-sonnet-4`).
+
+Each token setting resolves independently: matching model override → ordinary `compaction` setting → built-in default. In the example, `some-provider/big-model` keeps the ordinary 20000 recent tokens. Token values must be non-negative safe integers. Invalid values in the matching model override produce an error when read; only omitted fields fall back to the ordinary setting. Model override entries must be objects. Invalid ordinary token settings produce an error when read, even if the active model has a valid override. Only omitted ordinary values use built-in defaults. Zero is accepted, but `reserveTokens: 0` leaves no response margin and also sets the summarization output budget to zero.
+
+Global and project settings merge recursively **before** model lookup. A project can override one field for a model without replacing its other fields or other models. A global model-specific value takes precedence over a project-wide fallback; override the same model entry in the project to change it.
+
+`enabled` is not model-specific. The active model's token settings apply to manual compaction, automatic threshold checks (including between assistant turns), and overflow recovery. Switching models takes effect on the next check or compaction. Configure overrides in JSON; `/settings` retains the ordinary auto-compaction toggle.
+
+See [compaction.md](compaction.md) for trigger and summarization behavior.
+
 ### Branch Summary
 
 | Setting | Type | Default | Description |
@@ -143,10 +198,13 @@ Set `PI_SKIP_VERSION_CHECK=1` to disable the Pi version update check. Use `--off
 | `retry.enabled` | boolean | `true` | Enable automatic agent-level retry on transient errors |
 | `retry.maxRetries` | number | `3` | Maximum agent-level retry attempts |
 | `retry.baseDelayMs` | number | `2000` | Base delay for agent-level exponential backoff (2s, 4s, 8s) |
+| `retry.maxAgentDelayMs` | number | `60000` | Max agent-level retry delay (60s) |
 | `retry.provider.timeoutMs` | number | SDK default | Provider/SDK request timeout in milliseconds |
 | `retry.provider.maxRetries` | number | `0` | Provider/SDK retry attempts |
 | `retry.provider.maxRetryDelayMs` | number | `60000` | Max server-requested delay before failing (60s) |
 
+Agent-level retries use exponential backoff capped by `retry.maxAgentDelayMs`, so long retry runs stay responsive after prolonged outages.
+
 When a provider requests a retry delay longer than `retry.provider.maxRetryDelayMs`, the request fails immediately with an informative error instead of waiting silently. Set it to `0` to disable the limit.
 
 Keep `retry.provider.maxRetries` at `0` unless provider-level retries are explicitly needed. Setting it above `0` can make SDK/provider retries handle out-of-usage-limit errors before Pi sees them, which may block the agent until the provider quota resets in some circumstances.
@@ -157,6 +215,7 @@ Keep `retry.provider.maxRetries` at `0` unless provider-level retries are explic
     "enabled": true,
     "maxRetries": 3,
     "baseDelayMs": 2000,
+    "maxAgentDelayMs": 60000,
     "provider": {
       "timeoutMs": 3600000,
       "maxRetries": 0,

package/docs/termux.md

@@ -96,7 +96,6 @@ termux-camera-photo out.jpg   # Take photo
 ## Limitations
 
 - **No image clipboard**: Termux clipboard API only supports text
-- **No native binaries**: Some optional native dependencies (like the clipboard module) are unavailable on Android ARM64 and are skipped during installation
 - **Storage access**: To access files in `/storage/emulated/0` (Downloads, etc.), run `termux-setup-storage` once to grant permissions
 
 ## Troubleshooting